In [226]:
### Loading Libraries...
import pandas as pd
import numpy as np
### Graphic libraries
import matplotlib.pyplot as plt
import seaborn as sns
### Some Scikit-learn utils
from sklearn.model_selection import train_test_split
### Metrics
from sklearn import metrics
from sklearn.metrics import accuracy_score, roc_curve, auc
### Models
from xgboost import XGBClassifier, plot_importance
import itertools
from itertools import chain
from sklearn.feature_selection import RFE
from sklearn.decomposition import PCA
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import VotingClassifier
from sklearn.model_selection import GridSearchCV, cross_val_score, learning_curve, train_test_split
from sklearn.metrics import precision_score, recall_score, confusion_matrix, roc_curve, precision_recall_curve, accuracy_score
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.neural_network import MLPClassifier
from sklearn.metrics import f1_score,accuracy_score,roc_auc_score
import lime
from lime.lime_tabular import LimeTabularExplainer
import eli5
from eli5.sklearn import PermutationImportance
import shap
from shap import TreeExplainer,KernelExplainer,LinearExplainer
shap.initjs()
### Some cosmetics add-ons
import warnings
warnings.filterwarnings('ignore')
plt.style.use('fivethirtyeight')
%matplotlib inline
In [2]:
conda install -c plotly plotly
In [3]:
import plotly.offline as py
py.init_notebook_mode(connected=True)
import plotly.graph_objs as go
import plotly.tools as tls
import plotly.figure_factory as ff
In [246]:
data = pd.read_csv('/Users/odianosenakhibi/Downloads/Thesis/data/data.csv')
In [5]:
pip install PDPbox
In [247]:
np.random.seed(123) #ensure reproducibility
pd.options.mode.chained_assignment = None #hide any pandas warnings
In [248]:
from pdpbox import pdp, info_plots #for partial plots
In [249]:
import seaborn as sns
In [250]:
from xgboost import XGBClassifier, plot_importance
In [251]:
from sklearn import metrics
from sklearn.metrics import accuracy_score, roc_curve, auc
In [252]:
data.describe()
Out[252]:
In [253]:
# removing the last column as it is empty
data = data.drop(['id', 'Unnamed: 32'], axis = 1)
print(data.columns)
In [254]:
data.describe()
Out[254]:
In [255]:
data.isna().sum().sum()
Out[255]:
In [256]:
# Reassign target
data.diagnosis.replace(to_replace = dict(M = 1, B = 0), inplace = True)
In [257]:
# 2 datasets
M = data[(data['diagnosis'] != 0)]
B = data[(data['diagnosis'] == 0)]
In [258]:
plt.hist(data['diagnosis'])
plt.title('Diagnosis (M=1 , B=0)')
plt.show()
In [259]:
features_mean=list(data.columns[1:11])
# split dataframe into two based on diagnosis
dataM=data[data['diagnosis'] ==1]
dataB=data[data['diagnosis'] ==0]
In [260]:
#Stack the data
plt.rcParams.update({'font.size': 8})
fig, axes = plt.subplots(nrows=5, ncols=2, figsize=(8,10))
axes = axes.ravel()
for idx,ax in enumerate(axes):
ax.figure
binwidth= (max(data[features_mean[idx]]) - min(data[features_mean[idx]]))/50
ax.hist([dataM[features_mean[idx]],dataB[features_mean[idx]]], bins=np.arange(min(data[features_mean[idx]]), max(data[features_mean[idx]]) + binwidth, binwidth) , alpha=0.5,stacked=True, normed = True, label=['M','B'],color=['r','g'])
ax.legend(loc='upper right')
ax.set_title(features_mean[idx])
plt.tight_layout()
plt.show()
In [261]:
target = 'diagnosis'
features_list = list(data.columns)
features_list.remove(target)
In [262]:
#correlation
correlation = data.corr()
#tick labels
matrix_cols = correlation.columns.tolist()
#convert to array
corr_array = np.array(correlation)
In [263]:
# for visualizing correlations
#Plotting
trace = go.Heatmap(z = corr_array,
x = matrix_cols,
y = matrix_cols,
xgap = 2,
ygap = 2,
colorscale='Viridis',
colorbar = dict() ,
)
layout = go.Layout(dict(title = 'Correlation Matrix for variables',
autosize = False,
height = 720,
width = 800,
margin = dict(r = 0 ,l = 210,
t = 25,b = 210,
),
yaxis = dict(tickfont = dict(size = 9)),
xaxis = dict(tickfont = dict(size = 9)),
)
)
fig = go.Figure(data = [trace],layout = layout)
py.iplot(fig)
In [265]:
cat_cols = list(data.select_dtypes('object').columns)
class_dict = {}
for col in cat_cols:
data = pd.concat([data.drop(col, axis=1), pd.get_dummies(data[col])], axis=1)
data.head()
Out[265]:
In [266]:
# building train/test datasets on a 70/30 ratio
X_train, X_test, y_train, y_test = train_test_split(data, y, test_size=0.3, random_state=42)
X_train.shape, X_test.shape
Out[266]:
In [275]:
ML_models = {}
model_index = ['LR','RF','NN']
model_sklearn = [LogisticRegression(solver='liblinear',random_state=0),
RandomForestClassifier(n_estimators=100,random_state=0),
MLPClassifier([100]*5,early_stopping=True,learning_rate='adaptive',random_state=0)]
model_summary = []
for name,model in zip(model_index,model_sklearn):
ML_models[name] = model.fit(X_train,y_train)
preds = model.predict(X_test)
model_summary.append([name,f1_score(y_test,preds,average='weighted'),accuracy_score(y_test,preds),
roc_auc_score(y_test,model.predict_proba(X_test)[:,1])])
ML_models
Out[275]:
In [276]:
model_summary = pd.DataFrame(model_summary,columns=['Name','F1_score','Accuracy','AUC_ROC'])
model_summary = model_summary.reset_index()
display(model_summary)
In [269]:
g=sns.regplot(data=model_summary, x="index", y="AUC_ROC", fit_reg=False,
marker="o", color="black", scatter_kws={'s':500})
for i in range(0,model_summary.shape[0]):
g.text(model_summary.loc[i,'index'], model_summary.loc[i,'AUC_ROC']+0.02,
model_summary.loc[i,'Name'],
horizontalalignment='center',verticalalignment='top', size='large', color='black')
In [278]:
#initialization of a explainer from LIME
explainer = LimeTabularExplainer(X_train.values,
mode='classification',
feature_names=X_train.columns,
class_names=['Malign','Benign'])
In [279]:
exp = explainer.explain_instance(X_test.head(1).values[0],
ML_models['LR'].predict_proba,
num_features=X_train.shape[1])
exp.show_in_notebook(show_table=True, show_all=True)
In [280]:
exp = explainer.explain_instance(X_test.head(1).values[0],
ML_models['RF'].predict_proba,
num_features=X_train.shape[1])
exp.show_in_notebook(show_table=True, show_all=False)
In [281]:
exp = explainer.explain_instance(X_test.head(1).values[0],
ML_models['NN'].predict_proba,
num_features=X_train.shape[1])
exp.show_in_notebook(show_table=True, show_all=False)
In [231]:
#Eli5
eli5.show_weights(ML_models['LR'], feature_names = list(X_test.columns))
Out[231]:
In [277]:
eli5.show_prediction(ML_models['LR'], X_test.head(1).values[0],feature_names=list(X_test.columns))
Out[277]:
In [233]:
exp = PermutationImportance(ML_models['LR'],
random_state = 0).fit(X_test, y_test)
eli5.show_weights(exp,feature_names=list(X_test.columns))
Out[233]:
In [234]:
eli5.show_weights(ML_models['RF'],feature_names=list(X_test.columns))
Out[234]:
In [235]:
eli5.show_prediction(ML_models['RF'], X_test.head(1).values[0],feature_names=list(X_test.columns))
Out[235]:
In [236]:
exp = PermutationImportance(ML_models['RF'],
random_state = 0).fit(X_test, y_test)
eli5.show_weights(exp,feature_names=list(X_test.columns))
Out[236]:
In [237]:
eli5.show_weights(ML_models['NN'])
Out[237]:
In [238]:
#SHAP
explainer = LinearExplainer(ML_models['LR'], X_train, feature_dependence="independent")
shap_values = explainer.shap_values(X_test.head(1).values)
shap.force_plot(explainer.expected_value,
shap_values,
X_test.head(1).values,
feature_names=X_test.columns)
Out[238]:
In [239]:
shap_values = explainer.shap_values(X_test.head(250).values)
shap.force_plot(explainer.expected_value,
shap_values,
X_test.head(250).values,
feature_names=X_test.columns)
Out[239]:
In [240]:
shap_values = explainer.shap_values(X_test.values)
spplot = shap.summary_plot(shap_values, X_test.values, feature_names=X_test.columns)
In [242]:
top4_cols = ['area_worst','radius_worst','radius_mean','perimeter_worst']
for col in top4_cols:
shap.dependence_plot(col, shap_values, X_test)
In [243]:
explainer = TreeExplainer(ML_models['RF'])
shap_values = explainer.shap_values(X_test.head(1).values)
shap.force_plot(explainer.expected_value[1],
shap_values[1],
X_test.head(1).values,
feature_names=X_test.columns)
Out[243]:
In [245]:
X_train_kmeans = shap.kmeans(X_train, 10)
explainer = KernelExplainer(ML_models['NN'].predict_proba,X_train_kmeans)
shap_values = explainer.shap_values(X_test.head(1).values)
shap.force_plot(explainer.expected_value[1],
shap_values[1],
X_test.head(1).values,
feature_names=X_test.columns)
Out[245]:
In [ ]: